Birth data

Secondary data from non-centralized data source

Data from Danish midwife journals have been collected for all individuals enrolled in DD2 per 22 February 2022. Data include information on birth weight, birth length, born at term (yes/no, and if no then also the number of weeks), and twin (yes/no).

The “afleverings-/afslutningsrapport” from the Danish National Archive can be downloaded here (Danish):
A similar description in English is included in the electronic supplemental material from Hansen et al. (2023):

CPR numbers for a total of N=9,549 individuals were sent to the Danish National Archive and N=8,896 fulfilled the criteria for potentially having available birth information (born in Denmark in the period from ~1920 to 1988 and identifiable through the biological mother’s name), yet N=532 were not included despite fulfilling the criteria¹. During the retrieval, data were uniformized (e.g., to include birth weight in grams and not pounds) and proofread.

Control population

In addition to the individuals from DD2, a control population was recorded. For every individual in DD2, approximately two random individuals were selected based on the births from the same midwife information sheet (with birth information on ~6–8 different births). Controls from the same midwife record thus served as a match on date of birth, midwife, and geographical location. A total of N=18,210 individuals are recorded in the control population and data include information on year and month of birth, birth weight, birth length, sex, born at term, twin, and geographical location (lægekreds/fødselsamt). No CPR numbers are available for the control population (data permissions along with ethical and practical reasons) and neither are links between the controls and the individuals in DD2. Data on controls are not available on the servers at The Danish Health Data Authority (SDS) and Statistics Denmark (DST), as the data permissions only include individuals in DD2 and therefore not the matched controls. As CPR numbers are not available and controls ultimately cannot be linked to their respective DD2 individual there is no reason to upload data to the servers because all analyses can be performed locally.

Selected pages from the opgavebeskrivelse (del 1 and del 2) can be downloaded here (Danish):

Data documentation

foedselsdata.sas7bdat

Format (var x obs)	Id variables	Unique key	Important dates
Wide (9 x 9,544)	CPR	CPR	-

The dataset include birth data for N=9,544 individuals identified via CPR. There are no dates in the dataset, however, information relates to birth date. N=8,346 have non-missing birth weight, and a total of N=8,364 have a non-missing value in at least one of the five birth variables.

Illustration of the overall data structure. The dataset is in wide format (9 variables x 9,544 observations), with CPR as the unique key.
Row	CPR	foedselsvaegt	foedselslaengde	fuldbaarenhed	uger_for_tidligt	tvilllingefoedsel
1	CPR1	num.	num.	Ja		0
2	CPR2	num.	num.	Nej	2 til 3	0
3	CPR3	num.	num.	Ja		0
…	…	…	…	…	…	…
9,544	CPR9544	num.	num.	Ja		1

References

Hansen AL, Thomsen RW, Brøns C, Svane HML, Jensen RT, Andersen MK, et al. Birthweight is associated with clinical characteristics in people with recently diagnosed type 2 diabetes. Diabetologia. 2023;66(9):1680–92.

Footnotes

It is not marked in the data whether missing was due to not being eligible or due to other reasons↩︎